Hydromechanics neural network model evaluation method and device, electronic equipment and medium

Abstract

The invention discloses a fluid mechanics neural network model evaluation method and device, electronic equipment and a medium, and the method carries out the training and evaluation of each to-be-evaluated fluid mechanics neural network model through a sample data set corresponding to each fluid mechanics problem. And obtaining a target evaluation total value corresponding to each to-be-evaluated hydrodynamic neural network model under each sample data set, and determining an optimal hydrodynamic neural network model corresponding to the hydrodynamic problem from the plurality of to-be-evaluated hydrodynamic neural network models according to the target evaluation total value corresponding to each to-be-evaluated hydrodynamic neural network model. According to the method, the hydrodynamic neural network model evaluation can be realized, the performance, precision and universality of various hydrodynamic neural network models on various hydrodynamic problems are evaluated, the scientificity, systematicness and accuracy of hydrodynamic neural network model evaluation are improved, and the method can be widely applied to the technical field of model evaluation.

Description

Technical Field

[0001] The present application relates to the field of model evaluation technology, and in particular to a method, device, electronic equipment and medium for evaluating a fluid mechanics neural network model. Background Art

[0002] Computational fluid dynamics (CFD) has become an indispensable tool in the analysis and simulation of fluid flows in many fields such as aerospace, automotive engineering and environmental engineering. Traditional CFD methods usually rely on numerical solutions to the Navier-Stokes equations, which are computationally intensive and time-consuming when dealing with complex three-dimensional turbulent flows. As the demand for high-fidelity simulations increases, it becomes particularly important to develop more efficient computational methods.

[0003] In recent years, neural network (NN) methods have shown great potential in the field of CFD, which can significantly reduce computational costs while maintaining or even improving accuracy. These data-driven models can learn complex patterns and behaviors from large data sets, thereby quickly predicting fluid dynamics behavior without solving partial differential equations from scratch. This opens up new possibilities for real-time applications and large-scale simulations of CFD.

[0004] Despite rapid progress in this field, there is currently a lack of a systematic framework to evaluate and compare different neural network models for CFD applications. Summary of the invention

[0005] The main purpose of the embodiments of the present application is to propose a fluid mechanics neural network model evaluation method, device, electronic equipment and medium, which can realize fluid mechanics neural network model evaluation and evaluate the performance, accuracy and versatility of various fluid mechanics neural network models on various fluid dynamics problems.

[0006] On the one hand, the embodiment of the present application proposes a fluid mechanics neural network model evaluation method, the method comprising the following steps:

[0007] Obtain the physical system equation representation corresponding to each fluid mechanics problem;

[0008] Constructing sample data sets corresponding to each of the fluid mechanics problems using the equations of each of the physical systems;

[0009] Obtain multiple fluid mechanics neural network models to be evaluated;

[0010] For each of the fluid mechanics neural network models to be evaluated, using each of the sample data sets to train, infer and evaluate the fluid mechanics neural network model to be evaluated, obtain multiple evaluation indicators, and determine the evaluation indicator value corresponding to each of the evaluation indicators;

[0011] For the sample data sets corresponding to each of the fluid mechanics problems, determine the target evaluation total values ​​corresponding to each of the fluid mechanics neural network models to be evaluated based on the evaluation index values ​​corresponding to each of the fluid mechanics neural network models to be evaluated and the weight ratios corresponding to each of the evaluation indexes; and determine the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem from multiple fluid mechanics neural network models to be evaluated based on the target evaluation total values ​​corresponding to each of the fluid mechanics neural network models to be evaluated.

[0012] In some embodiments, constructing a sample data set corresponding to each of the fluid mechanics problems using each of the physical system equations specifically includes:

[0013] Obtaining related items corresponding to the physical system equation representation, wherein the related items at least include: equation expressions, equation physical parameters, initial and boundary value conditions, and domain geometry;

[0014] The related items are modified to generate a plurality of sample data.

[0015] In some embodiments, for each of the fluid mechanics neural network models to be evaluated, using each of the sample data sets to train, infer and evaluate the fluid mechanics neural network model to be evaluated, obtaining multiple evaluation indicators, and determining the evaluation indicator value corresponding to each of the evaluation indicators specifically includes:

[0016] Dividing the sample data set according to a preset ratio to obtain a training set, a validation set, and a test set;

[0017] Using the training set and the verification set to train and verify the fluid mechanics neural network model to be evaluated;

[0018] The test set is used to perform result inference on the trained fluid mechanics neural network model to be evaluated, and then the model is evaluated to obtain the evaluation index value corresponding to each of the evaluation indicators.

[0019] In some embodiments, for the sample data set corresponding to each of the fluid mechanics problems, according to each of the evaluation index values ​​corresponding to each of the fluid mechanics neural network models to be evaluated and the weight proportions corresponding to each of the evaluation indexes, the target evaluation total value corresponding to each of the fluid mechanics neural network models to be evaluated is determined, and according to the target evaluation total value corresponding to each of the fluid mechanics neural network models to be evaluated, the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem is determined from multiple fluid mechanics neural network models to be evaluated, specifically including:

[0020] For each of the fluid mechanics neural network models to be evaluated, sorting the evaluation index values ​​corresponding to the fluid mechanics neural network models to be evaluated to obtain a sorting result, assigning a value to each of the evaluation indicators according to the sorting result, and updating the evaluation index value corresponding to each of the evaluation indicators;

[0021] Obtain the score weight ratio corresponding to each of the evaluation indicators;

[0022] For each of the fluid mechanics neural network models to be evaluated, determining the target total evaluation value corresponding to the fluid mechanics neural network model to be evaluated according to the score weight ratio corresponding to each of the evaluation indicators and the updated evaluation indicator value;

[0023] For each of the fluid dynamics problems, based on the target total evaluation values ​​corresponding to each of the fluid mechanics neural network models to be evaluated, the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem is determined from multiple fluid mechanics neural network models to be evaluated, wherein the optimal fluid mechanics neural network model corresponds to the highest target total evaluation value.

[0024] In some embodiments, the target evaluation total value is calculated by the following formula:

[0025]

[0026] Among them, sum is the total value of the target evaluation, N is the number of evaluation indicators, ω i is the score weight ratio corresponding to the i-th evaluation indicator, A i is the evaluation index value corresponding to the i-th evaluation index.

[0027] In some embodiments, assigning a value to each of the evaluation indicators according to the ranking result and updating the evaluation indicator value corresponding to each of the evaluation indicators specifically includes:

[0028] Obtaining multiple rankings and ranking values ​​corresponding to each of the rankings;

[0029] According to the ranking results, determine the score ranking corresponding to each of the evaluation indicators;

[0030] For the score ranking rank corresponding to each of the evaluation indicators, a target ranking rank corresponding to the score ranking rank is determined from the multiple ranking ranks, and the evaluation indicator value is updated to the ranking value corresponding to the target ranking rank.

[0031] In some embodiments, the fluid mechanics problems include at least Darcy flow-related problems, Taylor-Green vortex-related problems, top cover-driven square cavity flow-related problems, pipe flow-related problems, cylinder flow-related problems, two-phase flow-related problems and three-dimensional periodic mountain flow-related problems; the evaluation indicators include accuracy indicators and efficiency indicators.

[0032] On the other hand, the embodiment of the present application provides a fluid mechanics neural network model evaluation device, the device comprising:

[0033] The first module is used to obtain the physical system equation representation corresponding to each fluid mechanics problem;

[0034] The second module is used to construct a sample data set corresponding to each of the fluid mechanics problems using the equation representation of each of the physical systems;

[0035] The third module is used to obtain multiple fluid mechanics neural network models to be evaluated;

[0036] The fourth module is used to train, infer and evaluate the fluid mechanics neural network model to be evaluated using each of the sample data sets, obtain multiple evaluation indicators, and determine the evaluation indicator value corresponding to each of the evaluation indicators;

[0037] The fifth module is used to determine the target total evaluation value corresponding to each fluid mechanics neural network model to be evaluated for the sample data set corresponding to each fluid mechanics problem, according to the evaluation index value corresponding to each fluid mechanics neural network model to be evaluated and the weight ratio corresponding to each evaluation index, and determine the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem from multiple fluid mechanics neural network models to be evaluated according to the target total evaluation value corresponding to each fluid mechanics neural network model to be evaluated.

[0038] On the other hand, an embodiment of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the fluid mechanics neural network model evaluation method described above when executing the computer program.

[0039] On the other hand, an embodiment of the present application proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the fluid mechanics neural network model evaluation method described above.

[0040] The embodiments of the present application include at least the following beneficial effects: the present application provides a fluid mechanics neural network model evaluation method, device, electronic device and medium, which train and evaluate each fluid mechanics neural network model to be evaluated through a sample data set corresponding to each fluid mechanics problem, obtain the target evaluation total value corresponding to each fluid mechanics neural network model to be evaluated under each sample data set, and determine the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem from multiple fluid mechanics neural network models to be evaluated according to the target evaluation total value corresponding to each fluid mechanics neural network model to be evaluated. The present application can realize fluid mechanics neural network model evaluation, evaluate the performance, accuracy and versatility of various fluid mechanics neural network models on various fluid dynamics problems, and improve the scientificity, systematicness and accuracy of fluid mechanics neural model evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flow chart of a fluid mechanics neural network model evaluation method provided in an embodiment of the present application;

[0042] Figure 2 It is a schematic diagram of the model evaluation in the embodiment of the present application;

[0043] Figure 3 It is a structural schematic diagram of a fluid mechanics neural network model evaluation device provided in an embodiment of the present application;

[0044] Figure 4 It is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0046] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0047] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0048] Reference Figure 1 , Figure 1 is an optional flow chart of a fluid mechanics neural network model evaluation method provided in an embodiment of the present application, the method may include but is not limited to steps S101 to S105:

[0049] Step S101, obtaining the physical system equation representation corresponding to each fluid mechanics problem;

[0050] Step S102, constructing a sample data set corresponding to each fluid mechanics problem using the equation representation of each physical system;

[0051] Step S103, obtaining a plurality of fluid mechanics neural network models to be evaluated;

[0052] Step S104, for each fluid mechanics neural network model to be evaluated, using each sample data set to train, infer and evaluate the fluid mechanics neural network model to be evaluated, obtain multiple evaluation indicators, and determine the evaluation indicator value corresponding to each evaluation indicator;

[0053] Step S105, for the sample data sets corresponding to the fluid mechanics problems, determine the target evaluation total values ​​corresponding to the fluid mechanics neural network models to be evaluated according to the evaluation index values ​​corresponding to the fluid mechanics neural network models to be evaluated and the weight ratios corresponding to the evaluation indexes, and determine the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem from multiple fluid mechanics neural network models to be evaluated according to the target evaluation total values ​​corresponding to the fluid mechanics neural network models to be evaluated.

[0054] In some embodiments, the fluid mechanics problems include at least Darcy flow-related problems, Taylor-Green vortex-related problems, top cover-driven square cavity flow-related problems, pipe flow-related problems, cylinder flow-related problems, two-phase flow-related problems and three-dimensional periodic mountain flow-related problems.

[0055] In some embodiments, step S102 may include but is not limited to steps S201 to S202:

[0056] Step S201, obtaining related items corresponding to the physical system equation representation, the related items at least including: equation expression, equation physical parameters, initial boundary conditions and domain geometry;

[0057] Step S202, modify the relevant items to generate multiple sample data.

[0058] In some embodiments, illustratively, physical parameters of equations in the physical system equation representation are modified to generate multiple sample data, initial and boundary conditions in the physical system equation representation are modified to generate multiple sample data, domain geometry corresponding to the physical system equation representation is modified to generate multiple sample data, and all sample data are combined to form a sample data set. Optionally, after modifying the relevant items corresponding to the physical system equation representation, high-precision numerical methods are used to discretize the equation expressions to generate multiple sample data. The high-precision numerical methods may include but are not limited to fluid mechanics numerical methods such as finite difference method, finite element method, boundary element method, finite volume method and finite analysis method.

[0059] For example, a dataset of the top cover driven square cavity flow problem is constructed, and Ω = (0, L) × (0, H) is a two-dimensional rectangular cavity, where L is the cavity width and H is the cavity height. The physical system equation represents the corresponding equation expression as follows:

[0060]

[0061] Among them, Re is the Reynolds number, u is the velocity field, p is the pressure, x and y are both spatial variables, and t is the time variable.

[0062] The initial boundary conditions are as follows:

[0063]

[0064] By changing the Reynolds number Re, boundary condition velocity u and the ratio of regional shape H / L in the physical system equation representation, multiple different types of sample data corresponding to the top cover driven square cavity flow problem can be generated respectively, and all the sample data are combined to obtain the sample data set of the top cover driven square cavity flow problem.

[0065] In some embodiments, step S104 may include but is not limited to steps S301 to S303:

[0066] Step S301, dividing the sample data set according to a preset ratio to obtain a training set, a validation set and a test set;

[0067] Step S302, using the training set and the validation set to train and validate the fluid mechanics neural network model to be evaluated;

[0068] Step S303, using the test set to perform result inference on the trained fluid mechanics neural network model to be evaluated, and then perform model evaluation to obtain evaluation index values ​​corresponding to each evaluation index.

[0069] In some embodiments, each sample data set is randomly divided into a training set, a validation set, and a test set by a preset ratio, such as 8:1:1.

[0070] In some embodiments, multiple fluid mechanics neural network models to be evaluated are obtained, including at least FNO, DeepONet, U-Net, U-NO, MPNN, NUFNO, GFormer, OFormer, GNOT, LSM, KNO, UPS, etc.

[0071] Optionally, model-related hyperparameters of each fluid mechanics neural network model to be evaluated are set, and then the training set is input into each fluid mechanics neural network model to be evaluated, and then the validation set is used to perform model verification on each trained fluid mechanics neural network model to be evaluated. Based on the calculation effect of the fluid dynamics problem of each fluid mechanics neural network model to be evaluated on the validation set, the hyperparameters of each fluid mechanics neural network model to be evaluated are optimized, so that the performance of each fluid mechanics neural network model to be evaluated is optimized, and the optimal model of each fluid mechanics neural network model to be evaluated based on the validation set during the model training process is obtained.

[0072] In some embodiments, the evaluation indicators include accuracy indicators and efficiency indicators. The accuracy indicators include at least MAE (mean absolute error), RMSE (root mean square error), MSE (mean square error), L2RE (L2 relative error), MaxError (maximum error) and NMSE (standard mean square error), and the efficiency indicators include Inference-time (inference time).

[0073] Each test set is used to perform model evaluation on each fluid mechanics neural network model to be evaluated, and the evaluation index value corresponding to each evaluation index is obtained. According to the ranking rank and score weight corresponding to the evaluation index value, the target evaluation value of each fluid mechanics neural network model to be evaluated under each test set is determined.

[0074] For example, refer to Figure 2 , assuming that there are fluid mechanics problems A to C, corresponding to test sets A to C, and fluid mechanics neural network models 1 to 4 to be evaluated, use each test set to evaluate each fluid mechanics neural network model to be evaluated, and obtain the target evaluation value A1 of the fluid mechanics neural network model 1 to be evaluated under the test set A, the target evaluation value A2 of the fluid mechanics neural network model 2 to be evaluated under the test set A, the target evaluation value A3 of the fluid mechanics neural network model 3 to be evaluated under the test set A, and the target evaluation value A4 of the fluid mechanics neural network model 4 to be evaluated under the test set A, and so on.

[0075] In some embodiments, step S105 may include but is not limited to steps S401 to S404:

[0076] Step S401, for each fluid mechanics neural network model to be evaluated, sorting the evaluation index values ​​corresponding to the fluid mechanics neural network model to be evaluated, obtaining the sorting result, assigning a value to each evaluation index according to the sorting result, and updating the evaluation index value corresponding to each evaluation index;

[0077] Step S402, obtaining the score weight ratio corresponding to each evaluation indicator;

[0078] Step S403, for each fluid mechanics neural network model to be evaluated, according to the score weight proportion corresponding to each evaluation index and the updated evaluation index value, determine the target evaluation total value corresponding to the fluid mechanics neural network model to be evaluated;

[0079] Step S404, for each fluid dynamics problem, according to the target evaluation total value corresponding to each fluid dynamics neural network model to be evaluated, determine the optimal fluid dynamics neural network model corresponding to the fluid dynamics problem from multiple fluid dynamics neural network models to be evaluated, where the optimal fluid dynamics neural network model has the highest target evaluation total value.

[0080] In some embodiments, step S401 may include but is not limited to steps S501 to S503:

[0081] Step S501, obtaining a plurality of rankings and ranking values ​​corresponding to each ranking;

[0082] Step S502, determining the score ranking corresponding to each evaluation indicator according to the ranking result;

[0083] Step S503 , for the score ranking rank corresponding to each evaluation indicator, determine a target ranking rank corresponding to the score ranking rank from multiple ranking ranks, and update the evaluation indicator value to the ranking value corresponding to the target ranking rank.

[0084] In some embodiments, optionally, assuming that there are 7 evaluation indicators, the fluid mechanics neural network model 1 to be evaluated is evaluated using the test set A to obtain evaluation indicator values ​​1 to 7, and each evaluation indicator value is sorted. The sorting results (the values ​​are sorted from low to high) are: evaluation indicator value 2, evaluation indicator value 3, evaluation indicator value 4, evaluation index value 5, evaluation index value 6, evaluation index value 1 and evaluation index value 7.

[0085] The ranking is assigned as 100 points for the first place, 90 points for the second place, 80 points for the third place, and 0 points for the rest. Then the evaluation index value 2 is updated to 100 points, the evaluation index value 3 is updated to 90 points, the evaluation index value 4 is updated to 80 points, and the remaining evaluation index values ​​are updated to 0 points.

[0086] In step S403 of some embodiments, based on the assigned evaluation index value and its corresponding score weight ratio, the target evaluation total value is calculated, and the target evaluation total value is calculated by the following formula:

[0087]

[0088] Among them, sum is the total value of the target evaluation, N is the number of evaluation indicators, ω i is the score weight ratio corresponding to the i-th evaluation indicator, A i is the evaluation index value corresponding to the i-th evaluation index.

[0089] Exemplarily, the total target evaluation value = indicator MAE score*10% + indicator RMSE score*10% + indicator MSE score*10% + indicator NMSE score*10% + indicator MaxError score*10% + indicator L2RE score*25% + indicator Inference-time score*25%. The score weight ratio can be adjusted according to actual needs and is not limited.

[0090] Reference Figure 3 , Figure 3 : is an optional structural schematic diagram of a fluid mechanics neural network model evaluation device provided in an embodiment of the present application, the device is used to implement the above-mentioned fluid mechanics neural network model evaluation method, and the device may include:

[0091] The first module is used to obtain the physical system equation representation corresponding to each fluid mechanics problem;

[0092] The second module is used to construct sample data sets corresponding to various fluid mechanics problems using the equations of various physical systems;

[0093] The third module is used to obtain multiple fluid mechanics neural network models to be evaluated;

[0094] The fourth module is used to train and evaluate the fluid mechanics neural network model to be evaluated using each sample data set, obtain multiple evaluation indicators, and determine the evaluation indicator value corresponding to each evaluation indicator;

[0095] The fifth module is used to determine the target total evaluation value corresponding to each fluid mechanics neural network model to be evaluated for the sample data sets corresponding to each fluid mechanics problem, according to the evaluation index values ​​corresponding to each fluid mechanics neural network model to be evaluated and the weight ratio of each evaluation index, and determine the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem from multiple fluid mechanics neural network models to be evaluated according to the target total evaluation value corresponding to each fluid mechanics neural network model to be evaluated.

[0096] The specific implementation of the fluid mechanics neural network model evaluation device is basically the same as the specific implementation example of the above-mentioned fluid mechanics neural network model evaluation method, and will not be repeated here.

[0097] The embodiment of the present application also provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned fluid mechanics neural network model evaluation method when executing the computer program. The electronic device can be any smart terminal including a tablet computer.

[0098] See also Figure 4 , Figure 4 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0099] The processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), a processor that is good at parallel processing of graphics and general computing tasks (Graphics Processing Unit), or one or more integrated circuits, etc., and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0100] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 902, and the processor 901 calls and executes the fluid mechanics neural network model evaluation method of the embodiment of this application;

[0101] Input / output interface 903, used to implement information input and output;

[0102] Communication interface 904, used to realize communication interaction between the device and other devices, which can be realized by wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.);

[0103] A bus 905 that transmits information between the various components of the device (e.g., the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);

[0104] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .

[0105] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned fluid mechanics neural network model evaluation method is implemented.

[0106] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0107] The embodiments of the present application provide a fluid mechanics neural network model evaluation method, device, electronic device and medium, which train and evaluate each fluid mechanics neural network model to be evaluated through a sample data set corresponding to each fluid mechanics problem, obtain the target evaluation total value corresponding to each fluid mechanics neural network model to be evaluated under each sample data set, and determine the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem from multiple fluid mechanics neural network models to be evaluated according to the target evaluation total value corresponding to each fluid mechanics neural network model to be evaluated. The embodiments of the present application can realize fluid mechanics neural network model evaluation, evaluate the performance, accuracy and versatility of various fluid mechanics neural network models on various fluid dynamics problems, and improve the scientificity, systematicness and accuracy of fluid mechanics neural network model evaluation.

[0108] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0109] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0110] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0111] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0112] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0113] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0114] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0115] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0116] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0117] It should be appreciated that embodiments of the present invention may be implemented or enforced by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable memory. The method may be implemented in a computer program using standard programming techniques including a non-transitory computer-readable storage medium configured with a computer program, wherein the storage medium so configured causes the computer to operate in a specific and predefined manner - according to the methods and drawings described in the specific embodiments. Each program may be implemented in a high-level procedural or object-oriented programming language to communicate with a computer system. However, if desired, the program may be implemented in an assembly or machine language. In any case, the language may be a compiled or interpreted language. In addition, the program may be run on a programmed ASIC for this purpose.

[0118] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0119] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.

Claims

1. A fluid mechanics neural network model evaluation method, characterized in that: The method comprises the following steps: Obtain the physical system equation representation corresponding to each fluid mechanics problem; Constructing sample data sets corresponding to each of the fluid mechanics problems using the equations of each of the physical systems; Obtain multiple fluid mechanics neural network models to be evaluated; For each of the fluid mechanics neural network models to be evaluated, using each of the sample data sets to train, infer and evaluate the fluid mechanics neural network model to be evaluated, obtain multiple evaluation indicators, and determine the evaluation indicator value corresponding to each of the evaluation indicators; For the sample data sets corresponding to each of the fluid mechanics problems, determine the target evaluation total values ​​corresponding to each of the fluid mechanics neural network models to be evaluated based on the evaluation index values ​​corresponding to each of the fluid mechanics neural network models to be evaluated and the weight ratios corresponding to each of the evaluation indexes; and determine the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem from multiple fluid mechanics neural network models to be evaluated based on the target evaluation total values ​​corresponding to each of the fluid mechanics neural network models to be evaluated.

2. The method according to claim 1, characterized in that: The constructing of sample data sets corresponding to the fluid mechanics problems by using the physical system equations specifically includes: Obtaining related items corresponding to the physical system equation representation, wherein the related items at least include: equation expressions, equation physical parameters, initial and boundary value conditions, and domain geometry; The related items are modified to generate a plurality of sample data.

3. The method according to claim 1, characterized in that The step of training, reasoning and evaluating the fluid mechanics neural network model to be evaluated using each of the sample data sets to obtain multiple evaluation indicators and determining the evaluation indicator value corresponding to each of the evaluation indicators specifically includes: Dividing the sample data set according to a preset ratio to obtain a training set, a validation set, and a test set; Using the training set and the verification set to train and verify the fluid mechanics neural network model to be evaluated; The test set is used to perform result inference on the trained fluid mechanics neural network model to be evaluated, and then the model is evaluated to obtain the evaluation index value corresponding to each of the evaluation indicators.

4. The method according to claim 1, characterized in that The sample data set corresponding to each of the fluid mechanics problems, according to each of the evaluation index values ​​corresponding to each of the fluid mechanics neural network models to be evaluated and the weight proportions corresponding to each of the evaluation indexes, determines the target evaluation total value corresponding to each of the fluid mechanics neural network models to be evaluated, and according to the target evaluation total value corresponding to each of the fluid mechanics neural network models to be evaluated, determines the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem from multiple fluid mechanics neural network models to be evaluated, specifically including: For each of the fluid mechanics neural network models to be evaluated, sorting the evaluation index values ​​corresponding to the fluid mechanics neural network models to be evaluated to obtain a sorting result, assigning a value to each of the evaluation indicators according to the sorting result, and updating the evaluation index value corresponding to each of the evaluation indicators; Obtain the score weight ratio corresponding to each of the evaluation indicators; For each of the fluid mechanics neural network models to be evaluated, determining the target total evaluation value corresponding to the fluid mechanics neural network model to be evaluated according to the score weight ratio corresponding to each of the evaluation indicators and the updated evaluation indicator value; For each of the fluid dynamics problems, based on the target total evaluation values ​​corresponding to each of the fluid mechanics neural network models to be evaluated, the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem is determined from multiple fluid mechanics neural network models to be evaluated, wherein the optimal fluid mechanics neural network model corresponds to the highest target total evaluation value.

5. The method according to claim 4, characterized in that The target evaluation total value is calculated by the following formula: Among them, sum is the total value of the target evaluation, N is the number of evaluation indicators, ω i is the score weight ratio corresponding to the i-th evaluation indicator, A i is the evaluation index value corresponding to the i-th evaluation index.

6. The method according to claim 4, characterized in that The step of assigning a value to each evaluation indicator according to the ranking result and updating the evaluation indicator value corresponding to each evaluation indicator specifically includes: Obtaining multiple rankings and ranking values ​​corresponding to each of the rankings; According to the ranking results, determine the score ranking corresponding to each of the evaluation indicators; For the score ranking rank corresponding to each of the evaluation indicators, a target ranking rank corresponding to the score ranking rank is determined from the multiple ranking ranks, and the evaluation indicator value is updated to the ranking value corresponding to the target ranking rank.

7. The method according to claim 1, characterized in that The fluid mechanics problems include at least Darcy flow-related problems, Taylor-Green vortex-related problems, top cover-driven square cavity flow-related problems, pipe flow-related problems, cylinder flow-related problems, two-phase flow-related problems and three-dimensional periodic mountain flow-related problems; the evaluation indicators include accuracy indicators and efficiency indicators.

8. A fluid mechanics neural network model evaluation device, characterized in that: The device comprises: The first module is used to obtain the physical system equation representation corresponding to each fluid mechanics problem; The second module is used to construct a sample data set corresponding to each of the fluid mechanics problems using the equation representation of each of the physical systems; The third module is used to obtain multiple fluid mechanics neural network models to be evaluated; The fourth module is used to train, infer and evaluate the fluid mechanics neural network model to be evaluated using each of the sample data sets, obtain multiple evaluation indicators, and determine the evaluation indicator value corresponding to each of the evaluation indicators; The fifth module is used to determine the target total evaluation value corresponding to each fluid mechanics neural network model to be evaluated for the sample data set corresponding to each fluid mechanics problem, according to the evaluation index value corresponding to each fluid mechanics neural network model to be evaluated and the weight ratio corresponding to each evaluation index, and determine the optimal fluid mechanics neural network model corresponding to the fluid mechanics problem from multiple fluid mechanics neural network models to be evaluated according to the target total evaluation value corresponding to each fluid mechanics neural network model to be evaluated.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the fluid mechanics neural network model evaluation method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the fluid mechanics neural network model evaluation method according to any one of claims 1 to 7 is implemented.

Images